Mentra

Meet Mentra,
Your Mental Ally.

Record with ease,

Analyze with clarity

Project

Type

Team Case Study

Team Memeber

Soyeon Lee Product Designer

Scope

From initial concept to final prototype

Timeline

10 Weeks

Tool

Figma

My Role

Product design

UX/UI desgin

User interview

User Survey

Design system

Usability test

Design iterations

Overview

Keeping patients engaged in mental health treatment with AI and digital phenotyping

This project started from a simple question: why do so many patients in psychiatric care stop treatment on their own? Together with my teammate Soyeon, we identified three core pain points through surveys and interviews: logging their state felt tedious and progress was hard to see.

Designing easier tracking and clear insights for mental health users

We designed Mentra to remove the hassle of managing mental health while keeping the user in control. We automatically track the user’s status and use AI to make manual logging effortless, show changes in condition through easy-to-read graphs and clear AI insights. By simplifying these tasks, we help users stay confident and informed about their treatment journey.

Design Outcome

Solution 01

Sensor-powered auto-tracking
to minimise user's tracking burden

We designed auto-logging powered by digital phenotyping and wearable devices to track users’ behaviours and health patterns automatically. We ensured consistent data collection even without active user input, significantly reducing the daily burden of manual tracking.

See your condition at a glance with meaningful color chips

Reflect on your day through sleep, stress, focus, and your med routine.

Condition at a glance with meaningful color chips

Reflect on user’s day through sleep, stress, focus, and your med routine.

Raw data before exploring AI-driven insights.

Together with what user logs later. this tracked data becomes the base for personalised AI feedback.

Review your raw data before exploring AI-driven insights.

Together with what you log later. this tracked data becomes the base for personalised AI feedback.

Design Outcome

Solution 02

AI-Driven active logging to
simplify manual tracking

We designed an AI-supported logging system that learns user patterns to provide timely, personalised prompts. This approach preserves the user’s sense of control while making active tracking feel effortless through tailored, AI-guided questions.

Log how you feel through simple, personalized questions

Just tap through multiple choice, sliders, or even short text input when needed. In case of open input, generative AI understands and processes your response.

Enable users to log their feelings through simple, personalised questions

Users can log their state via multiple-choice, sliders, or short text inputs. Generative AI then processes any open-ended responses to ensure an intuitive and effortless tracking experience.

Complete daily questions to unlock insights

Users gain access to AI-driven insights only after completing their daily questions. This serves as a gentle nudge to encourage consistent engagement with the service.

Complete today’s questions to unlock your insights.

AI-drivened Insights are revealed only after you complete the daily questions — a gentle nudge to encourage ongoing engagement.

Design Outcome

Solution 03

Weekly overview at a glance with AI insights to monitor treatment effectiveness

AI analysis integrates all logged and tracked data into a clear weekly summary. This allows users to monitor their treatment progress and stay informed without feeling overwhelmed by data.

Intuitive scores with weekly performance comparisons

Visual overviews of medication, sleep, stress, and mood help users maintain motivation and track their recovery.

AI learns user's routine, gives encouragement and advice.

The system identifies hidden patterns and offers smart suggestions to guide users’ next steps.

Correlating stress, sleep, and mood with medication

Users can observe how emotional and physical shifts relate to their medication over time.

AI-Crafted Daily Briefing

These briefings help users understand their current state and the effectiveness of their medication.

Visualising negative emotional patterns via heatmap

Digital phenotyping detects negative language patterns to show if medication is effectively easing difficult emotions.

AI-detected symptom triggers with actionable details

The system highlights unnoticed symptoms and provides detailed information to assist in future decision-making.

Discover

Desk Research

People quit mental health treatment on their own because it's hard to see how if it works.

Depression, anxiety, and ADHD are increasingly recognized today, yet many people still discontinue treatment on their own. We wondered if social stereotypes or medication stigma could be the reason. That curiosity led us to study the real pain points and design a service to address them.

Depression, ADHD Rates Are Rising

Depressionin Select OECD Countries, 2019–2020

ADHDMedication Prescriptions, 2012-2022

Treatment Discontinuation Remains High

“Although up to 75% of patients discontinue antidepressant use within six months, others continue indefinitely

Pharmacologic Treatment of Depression, Am Fam Physician. 2023;107(2):173-181, 2023

“Recently, there has been media and public interest regarding discontinuation of antidepressant treatment, especially in primary care

“It is important to carefully explore patient fears, expectations and ideas about antidepressant discontinution

Stopping antidepressants or not, Austrian Journal of General Practice, 2025 10.31128/AJGP-09-23-6967

Discover

User Survey

Quantifying the Pain Points: 82% of users responded that tracking their state is helpful, but only 48% actually tried, and 42% of them quit because it was too difficult to sustain.

Participants: 23 individuals with a balanced distribution of treatment experience for depression, anxiety, and ADHD.

Methodology: A Structured Survey built around five core themes: medication adherence, symptom tracking, doctor communication, information seeking, and psychological barriers.

Goal: To quantitatively validate the prevalence of key pain points and identify how commonly these challenges occur within the broader user group.

User Survey

Skipping doses

77% struggled to stay consistent

sometimes forget to take medicine

56% experienced side effects

skip doeses due to physical discomfort

Tracking is useful yet hard to maintain

48% tracked meds or mood

Self-monitoring is growing but not habitual

73% used phone notes

Digital tools are common

82% said tracking helped

Better doctor communication

42% skipped tracking

Too much effort and lack of suitable tools

Reliance on unverified online information

87% relied on doctors

Trust in professionals remains high

78% searched online

Still search the internet and online communities for information

Discover

User Interview

Users lack proper tools to track their data and cannot easily see their treatment progress.

To uncover the causal link between the frustrations identified in our survey and the final decision to quit, we conducted in-depth interviews. We selected a subset of 5 participants for a comparative analysis to capture deep and raw narratives of their treatment experiences.

Participants: A subset of 5 participants selected for a comparative analysis between those who discontinued and those who persisted.

Methodology: Semi-structured Interviews focusing on open-ended questions to capture raw and unfiltered journey experiences.

Analysis: Utilized Interview Colour-Coding to cluster raw data into the key themes, specifically identifying the link between tracking, progress checking, and obtaining trustworthy information

Pain Point Labels

Forgot to take medication

Bothersome / Lack of interests

No optimal way to track it

Hard to realise the progres (both treatment and side effects)

Not easy to access a doctor

Lack of understanding of the illness and meds

Hard to talk to people about the illness and the treatment

Not wanting to be reliable

Doubt the medication due to side effects

Hard to describe or share the symptoms or changes

Financial burden of treatment

Struggling to give up bad habits

Not enough reliable, accurate information

Lack of emotional support

Difficulty tracking each medication due to multiple prescriptions

Not enough involvement from family/caregiver in treatment

Clustered Pain Points

Bothersome / Lack of interests

No optimal way to track it

Hard to realise the progress

Interview Colour-Coding

Define

Problem Definition
& Solution Hypotheses

Automatic tracking of physical states via digital phenotyping will remove the hassle of manual logging.

AI-guided prompts for subjective data like mood will make manual recording feel less like a chore.

Problem: Manual tracking is too tedious and annoying for users to maintain daily.

Visualising data trends with AI insights will give users confidence that their treatment is working.

Problem: Without seeing clear progress, users easily doubt the effectiveness of their treatment.

Define

Persona Development

Two Representative Personas: From Skeptical Patients to Highly Motivated Achievers

We defined two personas to understand how users navigate mental health treatment. Journey mapping revealed where motivation and trust break down, leading to insights for a clearer, more reassuring, and effortless experience. Color-coded pain points helped visualize emotional friction across each step of the journey.

Persona 01

uncertain · anxious · forgetful

Edward

22 | Male | University Student

Depression for 6 months

Anti-depressant for 3 months

Overview

Edward hesitantly begins his treatment, uncertain about its effectiveness. He wants to get better but struggles to accept his condition and stay consistent with medication.

Pain Points

Forgt to take medication

Too bothersome to log

No optimal way to track it

Hard to realise the changes

Expectation

Clear understanding of his condition and mental health

Any kind of help that can reassure him about the treatment

Minimal effort in tracking his treatment and side effects

User Journey Map

I doubt the treatment. Maybe there’s a reason depression still feels stigmatized.

I’ve let go of my misconceptions about depression. Seeing how I’m actually doing, it seems worth trying treatment for a while.

Persona 02

motivated · inconsistent · analytical

Hana

31 | Female | IT Marketing

Diagnosed ADHD 3 years ago

Taking meds inconsistently

Overview

Hana is motivated to manage her ADHD, and wants to track the effect of her medication, but struggles to maintain consistency due to the lack of proper tools.

Pain Points

Forgt to take medication

Too bothersome to log

No optimal way to track it

Hard to realise the changes

Expectation

Guide for easier and efficient recording

Visualisation of records

Reliable and detailed information on medications and ADHD

User Journey Map

I can’t tell if the medication is helping me, or if the side effects outweigh the benefits.

I can see a patterns in when side effects hit and track my progress over time. I can track my condition consistently

Ideate

Solution Ideation

How might we simplify the tracking process while letting users stay active and involved in their logs?

After sorting all brainstormed ideas into categories as a team, we identified the MVP features using a prioritisation matrix. We focused on features with 'High Urgency' that make users feel recording is effortless from the start, as well as 'High Impact' features that provide long-term support throughout their treatment journey.

Feature Brainstorming & Open Card Sorting

Impact x Urgency Matrix

Ideate

From Structure to Sketches

Streamlining workflows: strategic separation of "data recording" and "data analysis insight"

We split the features into two separate tabs to match how users work. By putting 'Data Logging' first and 'Insights' second, we made it natural for users to record data and then check the results.

Information Architecture

Crazy 8s sketching with data visualisation

We used Crazy 8s to quickly sketch the most intuitive layouts. For the Data Insight tab, however, we first studied various visualisation techniques. This preparation allowed us to sketch the most suitable graphs for each data type, even within the one-minute limit.

Crazy 8's Sketches

Wireframe

Prototype

Design System

Visual concept

The visual direction is minimal and calming, with warm tones and simple serif typography that enhances trust, well suited for a mental health app.

Scalable design system

We build a scalable design system based on atomic principles, starting with the smallest reusable components. To ensure an inclusive experience, we used color blindness simulations to select a primary palette that is accessible and clearly visible to users with color vision deficiencies.

Prototyping

We applied interactions to key components and refined them continuously based on usability test scenarios.

Prototype with interactions

Test

Usability Test

How we validated if logging was easy and AI insights were useful

We conducted UT with 20-30s users diagnosed with depression, anxiety, or ADHD. We designed a workflow that mirrored the actual user journey, from onboarding and daily logging to reviewing AI-generated insights. To optimize the user flow, we conducted A/B testing and decided to move the auto-tracked data from the 'Logging' tab to the 'Insight' tab. To quantify the result, we used SUS analysis, achieving a score of 85.0 (Grade A).

Participants valued tracking what’s usually hard to measure

During usability testing, participants expressed that they highly value tracking "invisible" data, such as ambiguous feelings that are difficult to measure. While explicit scores can sometimes cause stress, presenting these subtle trends in a relatable way helps users recognise their progress.

We found a 'blind spot' where constant tracking becomes a psychological burden for some mental states.

Some participants noted that users with severe depression may find any daily activity overwhelming, while those with anxiety might find constant tracking obsessive.

Despite our initial focus on users sceptical of treatment, we identified a 'blind spot' where the app’s requirements could conflict with the user's mental state, suggesting a need for even more flexible interaction models.

Test

Design Iterations

Prioritising user's mental model over system logic order

The auto-tracked state data was originally placed under Reflection → Manual Data Logging because it represented raw logs prior to AI processing.

However, user testing showed that users don’t distinguish between raw and AI-processed data. They see it as one continuous flow: viewing their state and understanding insights. For them, locating state data inside the Insights tab felt more intuitive and aligned with their mental model.

Once this pattern appeared in the first interview, we treated it as a meaningful signal, validated it through an A/B test in the next session, and then updated the design accordingly.

As-is

Auto-tracked data in „Reflection“

To-be

Auto-tracked data in „Insights“

People click on what looks visually heavier.

Users considered the overall condition score very important, so we placed a large score badge at the top of the Insights tab. In our usability test, all users tapped on the score expecting more detailed content or interaction.

Although detailed content is already displayed below, this behavior suggested that users intuitively perceive this prominent score badge as an interactive element. So, we added an overlay that appears on tap, summarizing the score and its key components in a more direct way. This small redundancy makes the interaction more intuitive and helps users immediately understand what their score means.

The score indicator users kept trying to tap

On Tab

Added overlay with detailed score information on tap

Prototype

Takeaways from Usability Testing

Autonomy Over Automation: Empowering Users Through Control

We introduced automation to minimise user effort and friction. However, testing revealed that users still seek a degree of control, such as the ability to interact with scores and explore underlying data

Giving them the chance to do something themselves can help them feel more in control and motivated, which automation alone can't provide.

What’s Clear to Me Isn’t Always Clear to Users

The more effort we put into a screen, the more intuitive it feels to us as designers, but users don’t always see it that way. That’s why user feedback is essential, even for the parts we feel most confident about.

Users are willing to share sensitive medical data if they immediately perceive the value of AI-driven insights.

While users initially felt cautious about privacy, they found it worthwhile once they saw how the 'Overview' and 'AI Insights' provided clear benefits. We learned that transparency in how data is used, combined with immediate utility, effectively lowers the psychological barrier to sharing personal information.

Mentra
New App Design